Deploy any MCP tool as a production service in 60 seconds
Turn command-line MCP tools into scalable HTTP services without rewriting a single line of code. Deploy once, scale automatically, pay nothing when idle.
Share AI tools across teams from a central platform. No more "works on my machine" - every MCP tool becomes a reliable, discoverable service.
The MCP ecosystem has powerful tools but inconsistent deployment patterns. NimbleTools Core provides a universal deployment layer that works with any MCP server - stdio or HTTP, Python or Go, simple or complex.
Prerequisites:
- Helm 3.0+ for package management
- k3d for local Kubernetes cluster (automatic setup)
Installation:
# One command gets you running (creates local cluster if needed)
curl -sSL https://raw.githubusercontent.com/NimbleBrainInc/nimbletools-core/refs/heads/main/install.sh | bashWhat just happened?
- ✅ Local Kubernetes cluster created (k3d-nimbletools-quickstart)
- ✅ NimbleTools operator deployed and running
- ✅ REST API available for service management
- ✅ Ready to deploy your first MCP service
Verify it worked:
# Switch to the cluster context
kubectl config use-context k3d-nimbletools-quickstart
# Set namespace for convenience
kubectl config set-context --current --namespace=nimbletools-system
# Check that all pods are running
kubectl get pods
# Expected output:
# NAME READY STATUS RESTARTS AGE
# nimbletools-core-control-plane-64b889fbdb-xxxxx 1/1 Running 0 2m
# nimbletools-core-control-plane-64b889fbdb-xxxxx 1/1 Running 0 2m
# nimbletools-core-operator-7bf4f9f667-xxxxx 1/1 Running 0 2m
# nimbletools-core-rbac-controller-56f67c95dc-xxxxx 1/1 Running 0 2mDeploy your first MCP service:
# Fetch the echo server definition from the registry
curl https://registry.nimbletools.ai/v0/servers/ai.nimblebrain%2Fecho > echo-server.json
# Create a workspace
curl -X POST http://api.nimbletools.dev/v1/workspaces \
-H "Content-Type: application/json" \
-d '{"name": "my-workspace", "description": "My first workspace"}'
# Deploy the echo server
curl -X POST http://api.nimbletools.dev/v1/workspaces/my-workspace/servers \
-H "Content-Type: application/json" \
-d @echo-server.json
# Verify the service is running
kubectl get pods -n ws-my-workspaceSuccess! 🎉 You now have a production MCP service that scales and integrates with any MCP client.
Model Context Protocol (MCP) lets AI assistants use external tools and data sources. Think of it as "plugins for AI" - but instead of being locked to one AI system, MCP tools work with Claude, ChatGPT, local models, and more.
What problems does MCP solve?
- AI assistants can access real-time data (not just training data)
- Tools can be shared across different AI platforms
- Developers can create once, deploy everywhere
Common MCP tools: File browsers, database clients, API integrations, calculators, code analyzers, system monitors.
Skip the intro - you know MCP tools are powerful but deployment is inconsistent. Every tool has different requirements, some are stdio-only, others HTTP-only, and scaling is a manual nightmare.
NimbleTools Core solves this by:
- Universal Deployment: stdio tools run via our adapter, HTTP tools run natively
- Auto-scaling: Scale from 0 to N replicas automatically
- Production Ready: Built-in monitoring, security, and reliability
👉 Next: Deploy your existing MCP tools
You understand the operational complexity of running services at scale. NimbleTools Core provides Kubernetes-native primitives for MCP service lifecycle management.
Architecture highlights:
- Custom MCPService CRD with full lifecycle management
- Kubernetes operator pattern for automated operations
- Scale-to-zero with automatic cold-start handling
- Multi-tenant workspace isolation
👉 Next: Architecture deep-dive
🚀 60-Second Deployment
From zero to production MCP runtime in under a minute
⚡ Zero-Waste Scaling
Automatically scale from 0 to N replicas based on demand - pay nothing when idle
🔧 Works With Any MCP Tool
Deploy stdio command-line tools and native HTTP services with the same interface
🗂️ Service Registry
Browse and deploy from curated collections of MCP services - no more hunting across repos
🖥️ Command-Line Interface
Full CLI for automation, CI/CD, and developer workflows (ntcli)
🏢 Team-Ready from Day One
Multi-workspace support, pluggable authentication, comprehensive observability
🔒 Production Security
Kubernetes-native RBAC, network policies, signed images, vulnerability scanning
🛠️ Developer Friendly
Local development with k3d, hot-reload, comprehensive examples and documentation
Turn service discovery from hunting to browsing
Instead of hunting for MCP services across GitHub repos and documentation, browse curated collections of ready-to-deploy services from the NimbleTools Registry.
# Browse all available services
curl https://registry.nimbletools.ai/v0/servers | jq '.servers[].name'
# Get details for a specific server
curl https://registry.nimbletools.ai/v0/servers/ai.nimblebrain%2Fecho
# Deploy any service instantly
curl https://registry.nimbletools.ai/v0/servers/ai.nimbletools%2Fecho > echo-server.json
curl -X POST http://api.nimbletools.dev/v1/workspaces/my-workspace/servers \
-H "Content-Type: application/json" \
-d @echo-server.jsonAvailable in the registry:
ai.nimbletools/echo- Testing and debugging toolai.nimbletools/finnhub- Financial market data APIai.nimbletools/github- GitHub integration (repos, PRs, issues)ai.nimbletools/postgres-mcp- PostgreSQL database accessai.nimbletools/tavily-mcp- Web search and extractionai.nimbletools/nationalparks-mcp- US National Parks data- And more...
Registry features:
- Tested Services: Every service is pre-configured and tested
- Best Practices: Proper security, scaling, and monitoring built-in
- API Access: Browse and query servers programmatically
- Full Documentation: Complete API docs at registry.nimbletools.ai/docs
For Teams & Organizations: Publish your custom MCP services to your own registry or contribute to the community registry.
See the full documentation at registry.nimbletools.ai/docs for:
- Server definition schema
- Publishing requirements
- Validation and testing
- Registry API reference
Use cases:
- Enterprise Service Catalogs: Curated, approved services for your organization
- Team-Specific Tools: Department-specific MCP services and configurations
- Compliance & Security: Services that meet your organization's security requirements
- Custom Business Logic: Proprietary MCP services for internal processes
Learn more: NimbleTools Registry | API Documentation | GitHub Repository
Manage your MCP platform without leaving the terminal
Everything you can do through the web API, now available as intuitive CLI commands for automation and developer workflows.
Installation:
# Install ntcli
pip install ntcli
# OR
curl -sSL https://github.com/NimbleBrainInc/ntcli/releases/latest/download/install.sh | bashUsage:
# Configure your NimbleTools Core endpoint
ntcli config set endpoint http://api.nimbletools.dev
# Workspace management
ntcli workspace create dev-team
ntcli workspace list
# Service deployment
ntcli server deploy calculator --workspace prod
ntcli server scale calculator --replicas 5
# Registry integration
ntcli registry add https://my-company.com/registry.yaml
ntcli registry search databasePerfect for:
- CI/CD Pipelines: Automated deployments and infrastructure-as-code
- Developer Workflows: Quick service management from the terminal
- Operations Teams: Bulk operations and scripting
Learn more: GitHub Repository
Uses Helm charts to deploy the platform:
curl -sSL https://raw.githubusercontent.com/NimbleBrainInc/nimbletools-core/refs/heads/main/install.sh | bash# Clone for customization
git clone https://github.com/NimbleBrainInc/nimbletools-core.git
cd nimbletools-core
# Install with custom Helm values
./install.sh --namespace my-namespace --ingress-enabled --domain my-company.com# Set up complete development environment
./scripts/dev-setup.shOnce installed, access the management API:
# List workspaces
curl http://api.nimbletools.dev/api/v1/workspaces
# List services
curl http://api.nimbletools.dev/api/v1/workspaces/default/servers
# API documentation
open http://api.nimbletools.dev/docs# Check operator health
kubectl logs -l app.kubernetes.io/component=operator -n nimbletools-core-system
# Check service status
kubectl get mcpservices --all-namespaces
# Verify API health
curl http://api.nimbletools.ai/health"Cannot connect to Kubernetes cluster"
# Create local cluster first
k3d cluster create nimbletools-test --wait
./install.sh"MCPService not creating pods"
# Check operator logs for detailed error messages
kubectl logs -l app.kubernetes.io/component=operator -n nimbletools-core-system -f"Service not accessible"
# Port-forward and test directly
kubectl port-forward service/your-service 8080:8000
curl http://api.nimbletools.ai/health📚 Full Documentation - Complete guides and references
We welcome contributions! Whether you're fixing bugs, adding features, or improving documentation:
git clone https://github.com/NimbleBrainInc/nimbletools-core.git
cd nimbletools-core
./redeploy.shApache 2.0 - see LICENSE for details.
Ready to get started? Run the 60-second install and deploy your first MCP service:
curl -sSL https://raw.githubusercontent.com/NimbleBrainInc/nimbletools-core/refs/heads/main/install.sh | bashJoin our Discord community to connect with other contributors and maintainers.